This is awesome, but I don’t know enough about NNCP to understand how it ties in with sneakernets concretely. Obviously you are using a sneakernet for the routing/transport layers, but what I don’t understand is where the boundary of this abstraction is.
In other words, once I get a physical device with a “packet” on it, how much do I need to do? Do I just plug it in and NNCP detects it and puts the data where I want it to go? It seems like that’s what procrxpath is doing, but then it just sticks the incoming data in a tmp directory?? If this script is essentially a proxy for the overall server/sneakernet system that implements reliable transport (if that is an appropriateway to think about it), how do I write a client that puts the response data somewhere sensible? Presumably NNCP has some kind of hook I can use that says “hey, there’s new data and it lives at $DIR”?